MODAL SPACE - IN OUR OWN LITTLE WORLD by Pete Avitabile

Is there a difference between

Well . . . Let's explain what the
a roving hammer and roving
it depends
accelerometer modal test? differences could be

Illustration by Mike Avitabile

Is there a difference between a roving hammer and roving accelerometer test?

Well ... it depends
Let's explain what the differences could be.

Basically, there is no difference between a roving hammer and One way to run the test is to place the tri-axial accelerometer at
roving accelerometer modal test. This is true providing the a fixed location and impact, in one direction, at all 9 points. We
same measurements are collected. Let me explain by discussing would then obtain 27 FRFs for the structure. Another way to
this seemingly simple but tricky fine point about a modal test. run the test is to impact at one point, in one direction, and have
the tri-axial accelerometer rove to all 9 points. Again we would
Back when we performed a modal test with a 2 channel collect 27 FRFs. So in both cases, we measure 27 FRFs by
analyzer, it was fairly straightforward to perform an impact test. impacting in only one direction.
Usually, the hammer roved around the structure with a
stationary accelerometer. Typically, we impacted the structure But are the two tests the same? At first glance, you would
at every point in the x, y, and z directions to obtain FRFs think that both test setups should produce the same results. In
relative to the reference location of the stationary accelerometer. order to confirm whether this is true or not, let's step through the
But when we started using multichannel analyzers to perform measurement process and list out what measurements are
the same test, there are some slight differences that need to be actually being made for each test setup.
addressed. Let's consider an impact test for the 9 points shown
on the structure. Let's also assume that I have an impact Test Setup #1
hammer and a tri-axial accelerometer with a 4 channel FFT
analyzer or acquisition system. Let's say that we want to run a modal test shown in setup #1. In
this test, the tri-axial accelerometer is stationary at point 9 and
MODAL TEST CONFIGURATION measures x, y, and z outputs. The input hammer force is
applied in the z direction only and roves to each of the 9 points
z shown.
x y HAMMER
Now let's list each of the FRFs that will be collected from this
TRI-AXIAL test setup. When we impact point 1 in the z direction, the
ACCELEROMETER
8 4 response is measured at 9x, 9y, and 9z. So the FRFs measured
1 are 9x/1z, 9y/1z, 9z/1z for the first measurement made. Next
7
5 we impact point 2 in the z direction and the response is
2 measured at 9x, 9y, and 9z. This set of FRFs are 9x/2z, 9y/2z,
9
9z/2z. We can continue on here but I think you get the hang of
6 STRUCTURE
it. But what did we actually measure? Let's arrange all of these
3
measurements in the FRF matrix to see what we have.

_________________________________________________________________________________________________________________________
Is there a difference between a roving hammer and roving accelerometer test? Copyright 1998
SEM Experimental Techniques - Aug1998 Page 1 Pete Avitabile
 SETUP 1 SETUP 2
ROVING IMPACT ROVING TRI-AX
IN Z ONLY
STATIONARY
z
STATIONARY
TRI-AX AT 9 8 4 IMPACT AT 9 8 4 x
1 IN Z ONLY 1 y
z 7 7
y 5 5
x 2 2
9 9
6 1x
6 1x

3 1y 3 1y

1z 1z

2x 2x

2y 2y

THREE 2z 2z

PARTIAL 3x 3x
ONE
ROWS 3y 3y
FULL
OF 3z 3z
COLUMN
FRFS
OF
9x
FRFS 9x

9y 9y

9z 9z

1x 1y 1z 2x 2y 2z 3x 3y 3z 9x 9y 9z 1x 1y 1z 2x 2y 2z 3x 3y 3z 9x 9y 9z

When we take a close look at the FRF matrix, we notice that we
have measured only parts of three different rows of this matrix.
So we only have three partial descriptions of the characteristic
of the system. But in each of the partial descriptions, we can
only see the characteristic information in the z direction. This
would be fine if there was only motion in the z direction. But
what if there was significant motion in the z direction when the
structure is excited in the x direction? We have only measured
response due to excitation in the z direction!
Test Setup #2
Now let's say that we also want to run the modal test shown in
setup #2. In this test, the hammer impacts only in the z
direction at point 9. The tri-axial accelerometer roves to each of
the 9 points shown for this test, measuring the x, y, and z
directions.
Let's list each of the FRFs that will be collected from this test
setup. When we impact point 9 in the z direction, the response
is measured at 1x, 1y, and 1z. So the FRFs measured are 1x/9z,
1y/9z, 1z/9z for the first measurement made. Next we move the
accelerometer to point 2 and the response is measured at 2x, 2y,
and 2z. This set of FRFs are 2x/9z, 2y/9z, 2z/9z. So what did
we actually measure? Again, let's arrange all of these
measurements in the FRF matrix to see what we have.
Now we notice that we have measured one complete column of
the FRF matrix. Now we can describe the response of the
system in a more complete sense. We have now measured
enough FRFs that we can describe the response of the system
for all points. Of course, I'm assuming that the reference
location at point 9 in the z direction is not the node of a mode!
So what should I do?
So while it appeared on the surface that both tests were the
same, there actually is a difference!!! So how could I change
these test setups so that the same data is measured. Well, there
are two ways. First, Setup #1 could be changed as follows.
Instead of using a tri-axial accelerometer, we could use a single
uniaxial accelerometer to acquire data at 9z, for instance. But
the difference would be that the impact excitation needs to be
applied in the x, y and z directions. Then the data collected
would be a row of the FRF matrix with 9z as the reference.
This is exactly the same data as collected in Setup#2 provided
that reciprocity holds true.
The other way to make sure that the same data is collected is as
follows. In Setup #1, the impact hammer needs to be used to
excite the x, y, and z direction. So the roving hammer needs to
impact in all three directions. In Setup #2, the stationary impact
at point 9 would need to be used to excite the structure in all
three directions. Both tests would then produce 3 complete
rows or columns of the FRF matrix.
Now you still may be a little confused by this. I know it's not
easy to comprehend the first time you hear it. The best way to
convince yourself is to write out all the FRF measurements that
you intend to collect to assure that at least one complete row or
one complete column of the FRF matrix is acquired.
I hope this simple explanation helps to clear up your question.
You need to carefully think about the measurements you are
going to make. Remember what I always say: "Thinking is not
optional!" If you have any more questions about modal
analysis, just ask me.

_________________________________________________________________________________________________________________________
Is there a difference between a roving hammer and roving accelerometer test? Copyright 1998
SEM Experimental Techniques - Aug1998 Page 2 Pete Avitabile